Jira Manager

Shorthand used below: $PY = <skill_dir>/.venv/bin/python, $S = <skill_dir>/scripts

Pre-flight (run once per session)

Run the preflight script before any other operation:

python3 <skill_dir>/scripts/jira_preflight.py

It validates the entire environment in a single pass:

1. Python 3.10+ is available
2. Virtual environment exists and dependencies are installed (markdown, markdownify)
3. Config file (.jira.json) is found and has required fields (jira_url, project_key)
4. Credentials — env vars are set (never prints values)
5. Connectivity — API is reachable, credentials are valid
6. Field discovery — warns if field_mappings or issue_types are empty

If the venv or dependencies are missing, the preflight tells you to run:

python3 <skill_dir>/scripts/jira_setup_env.py

To skip the connectivity check (e.g. on repeated calls):

python3 <skill_dir>/scripts/jira_preflight.py --skip-connectivity

Fetch tickets

# Single ticket
$PY $S/fetch_tickets.py --key PROJ-101 --format detail

# Multiple tickets by key (--key is repeatable, or use comma-separated --keys)
$PY $S/fetch_tickets.py --key PROJ-101 --key PROJ-102 --key PROJ-103 --format table
$PY $S/fetch_tickets.py --keys PROJ-101,PROJ-102,PROJ-103 --format table

# JQL search
$PY $S/fetch_tickets.py --jql "project=PROJ AND type=Story" --format table

# Filter (builds JQL automatically — quote values with parens in shell)
$PY $S/fetch_tickets.py --filter 'assignee=currentUser()' 'status=In Progress'

# Children of epic or story
$PY $S/fetch_tickets.py --children-of PROJ-100

# Board (active sprint only by default; add --board-all for everything)
$PY $S/fetch_tickets.py --board-id 123
$PY $S/fetch_tickets.py --board-id 123 --filter 'assignee=currentUser()'

# List boards
$PY $S/fetch_tickets.py --boards

Formats: table (default — shows key, type, SP, priority, status, sprint, summary), detail, json. Options: --max-results N (default 50), --no-convert (skip markup conversion), --include-remote-links (fetch remote issue links for detail/json output).

Filter values that are JQL functions (e.g. currentUser(), now(), startOfDay()) are passed through unquoted. Plain values are auto-quoted.

Shell quoting: Always single-quote filter values containing parentheses (e.g. 'assignee=currentUser()') to prevent shell expansion.

Create tickets

# Story under an epic
$PY $S/create_ticket.py --type story --summary "Title" --description "Desc" \
  --epic PROJ-100 --story-points 2 --priority High

# Subtask under a story
$PY $S/create_ticket.py --type sub-task --summary "Subtask title" --parent PROJ-101

# With assignee (use display name — automatically resolved to accountId on Jira Cloud)
$PY $S/create_ticket.py --type story --summary "Title" --assignee "Jane Smith" --parent PROJ-100

# With extra required fields (use --fields for raw JSON when named flags aren't enough)
$PY $S/create_ticket.py --type bug --summary "Bug title" \
  --fields '{"customfield_29823": {"value": "Dev"}, "customfield_29843": [{"value": "Production"}]}'

Create flags: --type, --summary, --description, --epic, --parent, --priority, --assignee, --component (repeatable), --fix-version (repeatable), --sprint, --labels, --story-points, --attachment (repeatable), --fields (raw JSON for custom fields), --copy-fields-from ISSUE-KEY (copy custom fields like QBR/team from an existing issue).

--copy-fields-from: Fetches all custom fields from the source issue and applies them to the new issue. Only copies customfield_* fields that aren't already set by other flags. Automatically skips non-copyable fields (e.g. GreenHopper rank fields) by checking field schema metadata. Useful when your Jira project has required custom fields (e.g. QBR, QBR Theme) that vary per team/project — just point at a sibling issue instead of hunting for field IDs.

Create order: epics → stories → subtasks. On failure, stop — do not continue with dependents.

If create fails with a 400 error, the script auto-diagnoses the missing field. Always follow the auto-diagnosis output:

• Fields marked ★ Recommended are on the create screen with allowed values — use the Fix: line verbatim.
• Fields marked "not on create screen" cannot be set during creation — skip them.
• If multiple matches appear, the recommended one is almost always correct. Do NOT try non-settable fields first.
• You can also run $PY $S/discover_fields.py --fields-for-type <type> manually to explore available fields.

Bulk create

$PY $S/bulk_create.py --source tickets.md --epic PROJ-100 --dry-run

Remove --dry-run after review. See SOURCE_FORMAT.md for format.

Update tickets

# Common fields
$PY $S/update_ticket.py --key PROJ-101 --summary "New Title" --story-points 3 --priority High

# Status transition
$PY $S/update_ticket.py --key PROJ-101 --status "In Progress"

# Assign by display name (auto-resolved to accountId on Jira Cloud)
$PY $S/update_ticket.py --key PROJ-101 --assignee "Jane Smith"

# Unassign (empty string = unassign)
$PY $S/update_ticket.py --key PROJ-101 --assignee ""

# Re-parent an issue (set parent to another issue)
$PY $S/update_ticket.py --key PROJ-101 --parent PROJ-200

# Any field by name (updates only — resolves via field_catalog)
$PY $S/update_ticket.py --key PROJ-101 --set "components=Backend"

# Comment, link, attachment (all combinable in one call)
$PY $S/update_ticket.py --key PROJ-101 --comment "Done." \
  --link "Blocks:PROJ-200" --attachment report.pdf

# Log work (time tracking)
$PY $S/update_ticket.py --key PROJ-101 --worklog "2h" --worklog-comment "Code review"
$PY $S/update_ticket.py --key PROJ-101 --worklog "1d"

Update flags: --summary, --description, --status, --priority, --assignee, --parent, --component, --fix-version, --sprint, --labels, --story-points, --set "field=value" (repeatable), --fields (raw JSON), --comment, --link "Type:KEY" (repeatable), --attachment (repeatable), --worklog "TIME" (e.g. 2h, 1d, 30m), --worklog-comment, --dry-run.

--parent: Sets the parent issue (e.g. --parent PROJ-200). Auto-wraps as {"key": "..."} for the API. If the update fails with a hierarchy error, the script auto-diagnoses the issue: shows both issue types' hierarchy levels and suggests what intermediate type to create. Also works via --set "parent=PROJ-200".

Bulk update

$PY $S/bulk_update.py --tickets "PROJ-1,PROJ-2" --status Done --confirm
$PY $S/bulk_update.py --board-id 123 --sprint "Sprint 5" --confirm
$PY $S/bulk_update.py --jql "project=PROJ AND priority=High" --priority Medium --confirm

Defaults to dry-run preview without --confirm.

Delete tickets

$PY $S/delete_ticket.py --key PROJ-110 --dry-run    # preview
$PY $S/delete_ticket.py --key PROJ-110 --confirm     # execute

Never delete without explicit user approval.

Manage comments

$PY $S/issue_comments.py list --key PROJ-101
$PY $S/issue_comments.py add --key PROJ-101 --body "Thanks for the update."
$PY $S/issue_comments.py edit --key PROJ-101 --comment-id 12345 --body "Revised note."
$PY $S/issue_comments.py delete --key PROJ-101 --comment-id 12345

List projects

$PY $S/list_projects.py

Discover fields

$PY $S/discover_fields.py --all --apply                # full discovery + save to config
$PY $S/discover_fields.py --search "QBR"               # find a field by name
$PY $S/discover_fields.py --fields-for-type epic        # list all fields + values for a type
$PY $S/discover_fields.py --transitions PROJ-101        # list available workflow transitions

Diff & validate

$PY $S/diff_tickets.py --manifest              # diff local vs Jira
$PY $S/validate_estimates.py --epic PROJ-100   # check sub-ticket estimate sums

Rules

1. ALWAYS use the provided scripts. NEVER write inline Python or call the Jira REST API directly.
2. If a script fails, debug the invocation (wrong flags, missing config, missing --apply). Do NOT abandon scripts and write custom code.
3. Never create/update/delete without showing the plan and getting user approval.
4. Never print credentials.
5. --set works on both create and update — resolves field names via field_catalog. For fields that --set can't resolve, use --fields '{"customfield_123": "val"}' (raw JSON).
6. Descriptions auto-convert Markdown ↔ Jira markup. Use --no-convert to skip. --rewrite-links rewrites relative markdown links to git browse URLs.
7. Images in descriptions are auto-attached. Local image references (![alt](path/to/image.png)) are detected, the paths are rewritten to basenames for Jira wiki markup (!image.png!), and the files are uploaded as attachments after issue creation/update. HTTP/HTTPS URLs are left as-is. For --description-file, paths resolve relative to the file's directory first, then fall back to CWD if not found; for --description, relative to CWD. Works in create_ticket, update_ticket, and bulk_create.
8. Mermaid diagrams in descriptions (```mermaid code blocks) are automatically rendered to PNG and attached. Requires mmdc (npm i -g @mermaid-js/mermaid-cli) or npx. If neither is available the blocks are left as code with a warning.

Error quick-ref